3030 Sidco Drive Nashville, TN 37204 Office 615.256.7114 ...
3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
-
Upload
iaset-journals -
Category
Documents
-
view
218 -
download
0
Transcript of 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
1/12
www.iaset.us [email protected]
AIR QUALITY MODELING AT SIDCO INDUSTRIAL ESTATE,
COIMBATORE USING ANN MODEL
R. CHANDRASEKARAN1, R. VIGNESH2& M. ISAAC SOLOMON JEBAMANI3
1Research Scholar, Department of Civil Engineering, Government College of Technology, Coimbatore, Tamil Nadu, India
2PG Student, Department of Civil Engineering, Government College of Technology, Coimbatore, Tamil Nadu, India
3Professor, Department of Civil Engineering, Government College of Technology, Coimbatore Tamil Nadu, India
ABSTRACT
This study has investigated the potential use of systematic approach to develop Artificial Neural Network (ANN)
predicting models for the concentration of pollutants at a specific area in SIDCO, Coimbatore. The goal was to determine
the concentration of PM25, PM10, and TSPM in the atmosphere according to their relationship with the month and pollutant
concentration. Four models were run using Artificial Neural Networks, in the first three models, Months (1-12) were
considered as input vectors and Concentrations of PM2.5, PM10 and TSPM were considered as targets separately in each
model. In the fourth model, Concentrations of particulate matter PM2.5, PM10and TSPM were considered as input vectors,
and months were considered as targets. Corresponding results were obtained for each model with R value ranges from
0.40302 to 0.9045.The models developed were reprogrammed and trained in such a way to predict the pollutant
concentration in a particular month.
KEYWORDS: Air Comprises, Mixture Contains a Group, Air Quality Modeling
INTRODUCTION
Air is one of the essential survival elements of the human life. Air comprises of mixture of gases which is used in
breathing and a lot of other activities. The mixture contains a group of gases of nearly constant concentrations and a group
with concentrations that are variable in both space and time. By volume, dry air contains 78.09% Nitrogen, 20.95%
Oxygen, 0.93% Argon, 0.039% Carbon di-oxide, and small amounts of other gases. Air also contains a variable amount of
water vapour, on average around 1%. Air plays a critical part in humans life, that one cannot live without it even for a few
minutes. It is necessary to keep the air breathable and safe.
Air Pollution is any undesirable change in the composition of air, or literally the contamination and deterioration
of the environment/nature by releasing undesirable constituents in the atmosphere, which are lethal to the human beings
and also various forms of life. It disturbs the stability and the ecological balance of the surroundings. It acts as a main
reason for various diseases, breath related disorders, lung disorders etc. Major pollution is man-made and few are naturally
occurring. Immense growth in the population and in turn phenomenal increase in vehicular traffic, and growth of industries
forms a major source of air pollution. Release of undesirable constituents from the industries and vehicles forms the
predominant source of pollution. These are termed as pollutants, which can be Particulate Matter PM10, PM2.5, Gaseous
pollutants like SO2, NO2, NH3and metals like Lead etc. Air comprises of these pollutants generally in certain limits, but
once when it is above the limit it causes air pollution. According to National Ambient Air Quality Standards, there are
International Journal of Civil
Engineering (IJCE)
ISSN(P): 2278-9987; ISSN(E): 2278-9995
Vol. 3, Issue 5, Sep 2014, 17-28
IASET
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
2/12
18 R. Chandrasekaran, R. Vignesh & M. Isaac Solomon Jebamani
Impact Factor (JCC): 2.6676 Index Copernicus Value (ICV): 3.0
certain tolerance limits for these pollutants, which should not be exceeded. In order to predict the Quality of an air, certain
index values can be referred and verified whether the air is clean. Air Quality Index (AQI) or Air Pollution Index is such
number which gives the quality of the air. The AQI is commonly used to indicate the severity of pollution.
MODELLING
A model may be defined as a representation of the reality. The particular representation used in any given case
can take a number of forms. Generally, in order to build something huge like a bridge or a fly over or a building, a
miniature version of the same will be made as to give a projection of how it will be in the future, it predicts the structure,
shape and look of it in a small form giving an insight of how it will be in the future. Similarly, this is applied to
mathematics also, this termed as Mathematical Modelling, will predict the future of something representing it in the form
of equations or any graphs or charts etc.,
This follows the principle of two variables, dependent and independent variables, i.e., the dependent variables
changes according to the independent variables. In the case of air pollution studies, the pollutant concentrations of oxides
of sulphur, oxides of nitrogen, particulate matter and meteorological data are the independent variables and AQI is the
dependent variable. So, modeling can be done between these and equations stating the behavior can be drawn. On the
whole, a mathematical model explains the behavior of something, for instance Air quality using mathematical concepts or
equations.
ARTIFICIAL NEURAL NETWORK
Artificial Neural Network is a tool which mimics the brain, uses the principles that is followed in the brain. It is
also called as connectionist architectures, computing paradigms that emulate the basic workings and learning rules of the
human brain. The brain is composed of approximately 1011 neurons, connected to roughly 103 other neurons by axons.
Neurons are the elements of our nervous system which transmits information from one part to other part. A neuron consists
of four elements Dendrites, Soma, Axon, and Synapses. The neurons are interconnected in the layers of the brain where it
transmits and deciphers information.
Figure 1: Neural Network Structure
Artificial neural networks consists of three layers generally, input layer, hidden layer and the output layer
(Figure. 1). The number of layers in the neural networks can vary from a single layer to multiple layers. The layer that gets
the inputs from the external environment is called the input layer. The network outputs are generated from the output layer.
The hidden layer in the middle consists of neurons, which forms a relationship between the given input and the target and
predicts an output, which is again modified using the back propagation algorithm until a satisfactory output is obtained.
Weights and bias are assigned which can be changed using trial and error until a satisfactory fit is obtained. Hidden layers
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
3/12
Air Quality Modeling at Sidco Industrial Estate, Coimbatore Using Ann Model 19
www.iaset.us [email protected]
are sometimes linked to a black box within which the input data are mapped into outputs utilizing suitable activation
functions. Different transfer functions can be used according to the type of data.
Neural networks are trained using various training algorithms where it uses back propagation to assign weights
and biases in order to train the network. Levenberg Marquardt back propagation algorithm is the mostly used training
function. It is also known as the damped least squares method, is used to solve non-linear least squares problems.
The primary application of the algorithm is in the least squares curve fitting problem. Various transfer functions are used in
the neural networks according to the type of the data used. Purelin, logsig, tansig and Gaussian are some of the functions
which are used as transfer functions.
OBJECTIVES OF THIS PROJECT
In this paper, emphasis is made on the prediction of levels of air pollution in the Tamil Nadu Small Industries
Development Corporation Limited (TANSIDCO) Industrial Cluster, Coimbatore based on the data available for the year
2012.
To develop models using artificial neural networks for the year 2012.
Model 1:To establish the relationship between PM 10, PM 2.5 and TSPM with the months for the year 2012.
Model 2:To establish the relationship between PM 25and months for the year 2012.
Model 3:To establish the relationship between PM 10and months for the year 2012.
Model 4:To establish the relationship between TSPM and months for the year 2012.
The models developed were to be reprogrammed and trained in such a way to predict the pollutant concentrations
in a particular month
PREDICTION OF PM10AND TSPM AIR POLLUTION PARAMETERS
Mouhammd Alkasassbeh et al., used Artificial Neural Networks to predict the concentrations of PMi0and TSPM
Air Pollution parameters. A data set collected at Al-Fuhais cement plant for over one-year period (2006, 2007) by eight
monitoring stations were used for this study. The ANN models used considered the meteorological parameters:
Temperature, Relative Humidity, Wind Speed as inputs and the targets were the concentrations of PMi0 and TSPM.
Two Artificial Neural Network based Auto Regressive with external (ANNARX) Input models were used to provide high
performance modeling for the PMi0 and the TSPM air pollution parameters. Experimental results showed that the Auto
Regressive models can provide good modeling results using a limited number of measurements.
ARTIFICIAL NEURAL NETWORKS FOR THE IDENTIFICATION OF UNKNOWN AIR
POLLUTION SOURCES
S.L. Reich et al., used Artificial Neural Networks which used pattern recognition to identify the unknown air
pollution parameters. The problem that is addressed in this paper is the apportionment of a small number of sources from a
data set of ambient concentrations of a given pollutant. Three layers feed-forward ANN trained with a back-propagation
algorithm were selected. Gaussian dispersion model is used to build the test case. A dataset of hourly meteorological
conditions and measured concentrations were taken as the inputs to the network that is wired to recover relevant emission
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
4/12
20 R. Chandrasekaran, R. Vignesh & M. Isaac Solomon Jebamani
Impact Factor (JCC): 2.6676 Index Copernicus Value (ICV): 3.0
parameters of unknown sources as outputs. The rest of the model data were corrupted adding noise to some meteorological
parameters to test the goodness of the method to recover the correct answer. The ANN was applied to predict the 24 hour
SO2concentrations and were compared with the measured concentrations.
Boznar et al., used neural networks to predict short term SO2concentrations in highly polluted industrial areas of
complex terrain around the Slovenian Thermal Power plant at Sostanj, India. Because the classical methods for air
pollution modeling, such as dispersion models, were not reliable in topography, neural networks were applied to predict
SO2 pollution. There were 37 input variables, including time of day, temperature, wind speed, wind direction, solar
radiation, SO2concentration, relative humidity and emissions. The results were very promising. A multi-layer perceptron
with sigmoid transfer function was trained with a back propagation algorithm. Separate networks for different locations
were trained because of the micro- climatological situations appearing in the region. The half-hourly concentrations of SO2
were used to train the network. Three stations were observed to examine the ability of the neural network. However, the
results for other months or seasons were not tested for the same area. In addition, no statistical validation tests were carried
out to observe the performance of the models.
ASSESSMENT AND PREDICTION OF TROPOSPHERIC OZONE CONCENTRATION LEVELS
USING ARTIFICIAL NEURAL NETWORKS
S.A. Abdul-Wahab et al., used Artificial Neural Networks to predict the concentration levels of ozone in the
troposphere. The network was trained using summer meteorological and air quality data where the ozone concentrations
were the highest. The data were collected from an urban atmosphere. Three neural network models were developed.
The first model was used on studying the factors that control the ozone concentrations during a 24-hour period where both
daylight and night hours were included. The second model was developed to study the factors that regulate the ozoneconcentrations during daylight hours at which higher concentrations of ozone were recorded. The third model was
developed to predict the daily maximum ozone levels. The predictions of the models were consistent with the measured
observations. A partitioning method was used to study the relative percent contribution of each of the input variables.
The contribution of meteorology on the ozone concentration variation was found to fall within the range 33.15-40.64%.
It was also found that Nitrogen oxide, Sulfur dioxide, relative humidity, non-methane hydrocarbon and Nitrogen dioxide
have the most effect on the predicted ozone concentrations. In addition, temperature played an important role while solar
radiation had a lower effect than expected. The results of this study indicate that the artificial neural network (ANN) is a
promising method for air pollution modeling.
FORECASTING AIR POLLUTION TIME SERIES USING NEURAL NETWORKS
Harri Niska et al., used neural networks to forecast the Air pollution time series. Genetic algorithm was used for
selecting the inputs and designing the high level architecture of a multi-layer perceptron model for forecasting hourly
concentrations of Nitrogen dioxide at a busy urban traffic station. The results showed considerable relevance between the
observed and the forecasted values. Only two hidden layers were required to get better results.
Benevenuto et al., illustrated the use and some related results of Artificial neural networks for data quality control
of environmental time series and for reconstruction of missing data. Artificial neural networks were applied to the short
and medium term prediction of air pollutant concentrations in urban areas, interpolating and extrapolating daily maximum
temperature, replacing time distribution with spatial distributions. Observed versus Predicted data were compared to test
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
5/12
Air Quality Modeling at Sidco Industrial Estate, Coimbatore Using Ann Model 21
www.iaset.us [email protected]
the efficacy of the ANNs in simulating the environmental processes. Their results confirmed ANNs as an improvement of
classical models and showed the utility of ANNs for restoration of time series.
OZONE AND PM10 FORECASTING
Ojha et al., presented a compendium of available methods and software for ozone and PM10 forecasting.
Three different methods of Regression analysis, time series analysis and artificial neural networks were discussed in this
paper. Lists of available software that can be used as a starting point, for the development of forecasting models were also
provided.
Gardner et al., did a case study with the UK data and demonstrated that statistical models of hourly surface ozone
concentrations require interactions and non-linear relationships between predictor variables in order to accurately capture
the ozone behavior. Comparisons between linear regressions, regression tree and multi-layer perceptron neural network
models of hourly surface ozone concentrations quantify these effects. They reported that although multi-layer perceptron
models were shown to more accurately capture the underlying relationship between both the meteorological and temporal
predictor variables and hourly ozone concentrations, the regression models were seen to be more readily physically
interpretable.
Asha Chelani et al., employed Artificial neural networks to predict the concentration of ambient respirable
particulate matter (PM10) and toxic metals observed in the city of Jaipur, India. A feed-forward network with a back
propagation learning algorithm was used to train the neural network to analyze the behavior of the data patterns.
The meteorological variables of wind speed, wind direction, relative humidity, temperature and time were taken as input to
the network. The results indicated that the network was able to predict the concentrations of PM10 and toxic metals
accurately.
ATMOSPHERIC DISPERSION IN COMPLEX TERRAIN
Sarkar and Jaleel presented a comparative study of the predictions of atmospheric dispersion in complex terrain
by conventional dispersion models and Artificial neural networks. A multi-layer feed forward back propagation with
generalized data rule applied for this study performed better than the mathematical models for all the test data.
PREDICTION OF GROUND LEVEL AIR POLLUTION
Mahad S. Baawain et al., used a statistical approach to predict the ground level air pollution around an industrial
port using Artificial neural networks. A rigorous method of preparing air quality data is proposed to achieve more accurate
air pollution prediction models based on artificial neural networks. The models consider the prediction of daily
concentrations of various ground level air pollutants, namely CO, PM10, NO2, NOx, H2S and Ozone, which were measured
by an ambient air quality monitoring station in Ghadafan village, located 700 m downwind of the emissions of Sohar
Industrial port on the Al-Batinah coast of Oman. The training of the models was based on the multi-layer perceptron
method with the back propagation algorithm. The results show very good agreement between the actual and predicted
concentrations with the R2value exceeding 0.70. The results also show the importance of temperature in daily variations of
ozone, SO2and NOx, while the wind speed and wind direction play important roles in the daily variations of NO, CO, NO 2
and H2S. PM10concentrations were influenced by almost all the measured meteorological parameters.
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
6/12
22 R. Chandrasekaran, R. Vignesh & M. Isaac Solomon Jebamani
Impact Factor (JCC): 2.6676 Index Copernicus Value (ICV): 3.0
SIDCO COIMBATORE STUDY AREA
Artificial Neural Network modeling was done for the Air Quality data which were sampled in
SIDCO- Coimbatore. The data consists of the concentrations of Particulate Matter PMi 0and PM25and Total Suspended
Particulate Matter, these were taken as inputs. On the other hand, months (112) were given as outputs. In Coimbatore
District, Kurichi is located at 105511 N latitude and 76 5735E longitude comprising of Industrial Cluster.
This Industrial cluster is located at distance of 7 km from Coimbatore Corporation. In Kurichi, two Industrial estates exist,
which are developed by SIDCO and Private. Adjoining to this estate, Tamilnadu Housing Board has constructed Housing
units. This Industrial cluster area spreads over an area of about 180 acres. This cluster comes under the administrative
jurisdiction of Kurichi Municipality. This industrial cluster is located on the National Highway from Coimbatore to
Pollachi.
SOFTWARE USED
MATLAB 7.12.0.635 (R2011a) software was used for carrying out the Artificial neural networks. Neural network
option in the software was used to run the model.
DATA
Ambient Air Quality data like concentrations of pollutants such as Particulate Matter PM10, PM2.5, and Total
Suspended Particulate Matter were obtained from previous experimental studies of Air Quality Monitoring at SIDCO for
the year 2012 by Er. R. Chandrasekaran, Assistant Environmental Engineer, Tamil Nadu Pollution Control Board,
Coimbatore.
Data preparation is one of the critical and the most important step while modeling through artificial neural
networks, because it has an immense impact on the success and the performance of the neural network results. The dataset
consisted of the concentrations of the Particulate matter PM10, PM25 and TSPM for the entire year of 2012. Data
preparation was done according to the model which is to be run. Four models were developed for establishing the
relationship between the PM10, PM25and TSPM with the months. In model I, PM10, PM25and TSPM were taken as inputs
and months were taken as outputs. In model II, PM25was taken as input and months were taken as outputs. In model III,
PM10was taken as input and months were taken as outputs. In model IV, TSPM was taken as input and months were taken
as outputs. Table 1 shows the input data and output data used for modeling evenberg-Marquardt back propagation
algorithm was used for training the model since it is been recommended as the first choice supervising algorithm. While
training with the help of this algorithm, the weights and biases are updated automatically by this algorithm.
Table 1: Concentrations of Pollutants and Months
Months PM2.5 (g/M3) PM10 (g/M3) TSPM (g/M3)
1 19 63 107
1 28 91 150
1 19 65 104
1 26 86 138
1 37 120 174
1 39 119 178
1 25 81 130
1 29 89 142
2 49 109 230
2 59 138 295
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
7/12
Air Quality Modeling at Sidco Industrial Estate, Coimbatore Using Ann Model 23
www.iaset.us [email protected]
Table 1: Contd.,
2 51 112 245
2 40 92 205
2 64 121 304
2 65 123 308
2 52 122 277
2 64 148 307
3 40 92 205
3 24 79 198
3 25 78 195
3 41 97 212
3 37 119 275
3 35 196 255
3 32 94 237
Levenberg-Marquardt back propagation algorithm was used for training the model since it is been recommended
as the first choice supervising algorithm. While training with the help of this algorithm, the weights and biases are updated
automatically by this algorithm. Ambient Air Quality data like concentrations of pollutants such as Particulate Matter
PM10, PM2.5, Total Suspended Particulate Matter and Months were used to run the Artificial neural network. Four models
were developed, in which the models used the concentrations of PM10, PM2.5 and TSPM as inputs and months as outputs.
Table 2: Predictor Variables Proposed for Each Model
Model No. Model Inputs Model Targets
I PM10, PM2.5, TSPM MonthsII PM2.5 MonthsIII PM10 Months
IV TSPM Months
MODEL I
The model was developed to establish a relationship between the months and the different concentrations of
PM1o, PM2.5, TSPM. The model was trained with different hidden layers to get considerable R2value. Figure 3 shows the
program used for model I. MODEL II The model was developed to establish a relationship between the months and the
different concentrations of PM2.5. Model was trained with different hidden layers to get a considerable R2value.
MODEL II
The model was developed to establish a relationship between the months and the different concentrations of
PM2.5. Model was trained with different hidden layers to get a considerable R2value.
MODEL III
This model was developed to establish a relationship between the months and the concentrations of PMi0.
The model was trained with different hidden layers to get a considerable R2value..
MODEL IV
The model was developed to establish a relationship between the months and the concentrations of TSPM.
The model was trained with different hidden layers to get a considerable R 2value. 16. VALIDATION OF THE MODEL
Ambient Air Quality data like concentrations of pollutants such as Particulate Matter PM10, PM25, Total Suspended
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
8/12
24 R. Chandrasekaran, R. Vignesh & M. Isaac Solomon Jebamani
Impact Factor (JCC): 2.6676 Index Copernicus Value (ICV): 3.0
Particulate Matter and Months were used to run the Artificial neural network. Four models were developed, in
Table 3: Predictor Variables for Each Model
Model No. Model Inputs Model Targets
I PM10, PM2.5, TSPM MonthsII PM2.5 Months
III PM10 Months
IV TSPM Months
which the models used the concentrations of PM10, PM25and TSPM as inputs and months as outputs.
PREDICTIONS FOR MODEL I
Artificial neural network model was run with the concentrations of particulate matter PM10, PM2.5, TSPM as input
and months as the output. Table 4 shows the best architectures for the model, where hidden layers ranging from 2 to 8 were
used and the corresponding R-Value was ranging from 0.70756 to 0.9045. The best R-Value was obtained at the hidden
layer of 8.
Figure 2: Neural Network Structure for Model I
Figure 3: Regression Plot for Model I
Table 4: Best Architectures for Model I
Hidden Layers R- Value
2 0.70756
3 0.72881
4 0.79868
5 0.79173
6 0.82897
7 0.88346
8 0.9045
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
9/12
Air Quality Modeling at Sidco Industrial Estate, Coimbatore Using Ann Model 25
www.iaset.us [email protected]
MODEL II
Artificial neural network model was run with the concentrations of PM2.5 as the input and months as the output.
Table 5 shows the best architectures for the model, where hidden layers ranging from 2 to 8 were used and the
corresponding R-Value was ranging from 0.45041 to 0.83846. The best R-Value was obtained at the hidden layer 5 to
Artificial neural network model was run with the concentrations of PM10 as input and months as output. Table 6 shows the
best architectures for the model, where hidden layers ranging from 2 to 8 were used and the corresponding R-Value was
ranging from 0.40302 to 0.67753. The best R-Value was obtained at the hidden layer of 4 to 8.
Figure 4: Neural Network Structure for Model II
Figure 5: Regression Plot for Model II
Table 5: Best Architectures for Model II
Hidden Layers R- Value
2 0.45051
3 0.469544 0.46954
5 0.83735
6 0.83735
7 0.83846
8 0.83846
PREDICTIONS FOR MODEL III
Artificial neural network model was run with the concentrations of PM10as input and months as output. Table 6
shows the best architectures for the model, where hidden layers ranging from 2 to 8 were used and the corresponding
R-Value was ranging from 0.40302 to 0.67753. The best R-Value was obtained at the hidden layer of 4 to 8.
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
10/12
26 R. Chandrasekaran, R. Vignesh & M. Isaac Solomon Jebamani
Impact Factor (JCC): 2.6676 Index Copernicus Value (ICV): 3.0
Figure 6: Neural Network Structure for Model III
Figure 7: Regression Plot for Model III
Table 6: Best Architectures for Model III
Hidden Layers R- Value
2 0.40302
3 0.67463
4 0.677535 0.67753
6 0.67753
7 0.67753
8 0.67753
PREDICTIONS FOR MODEL IV
Artificial neural network model was run with concentrations of TSPM as input and month as the output. Table 7
shows the best architectures for the model, where hidden layers ranging from 2 to 8 were used and the corresponding
R-Value was ranging from 0.80191 to 0.87119. The best R-Value was obtained at the hidden layer of 6 to 8.
Figure 8: Neural Network Structure for Model IV
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
11/12
Air Quality Modeling at Sidco Industrial Estate, Coimbatore Using Ann Model 27
www.iaset.us [email protected]
Figure 9: Regression Plot for Model IV
Table 7: Best Architectures for Model IV
Hidden Layers R- Value
2 0.80191
3 0.80191
4 0.87066
5 0.87066
6 0.87119
7 0.87119
8 0.87119
CONCLUSIONS
Air pollution models can be a very effective tool in planning strategies for management of local air quality and
can provide a rational basis for the control of air pollution. If properly designed and evaluated, air pollution models play a
considerable role in any air quality management system. In the present work, the most convincing advantage of ANN
model is that this model can be used in two ways, first we can predict the month for a particular concentration of PM 25,
PM10, and TSPM. Secondly we can predict the pollutant concentration based on the month.
REFERENCES
1. Meenakshi P., M. K. Saseetharan, September 2003, "Analysis of Seasonal Variation of Suspended Particulate
Matter and Oxides of Nitrogen with Reference to Wind Direction in Coimbatore City., IE (I) Journal.
EN Vol. 84.
2.
Daewon Byun and Kenneth L. Schere, September 2004, "Review of the Governing Equations, Computational
Algorithms, and Other Components of the Models-3 Community Multi scale Air Quality (CMAQ) Modeling
System Accepted by Applied Mechanics Reviews.
3. Alan J. Cimorelli, Steven G. Perry, Akula Venkatram, Jeffrey C. Weil, Robert J. Paine, Robert B. Wilson, Russell
F. Lee, Warren D. Peters, AND Roger W. Brode, October 2004 "AERMOD: A Dispersion Model for Industrial
Source Applications. Part I: General Model Formulation and Boundary Layer Characterization 682 Journal of
Applied Meteorology, Volume 44.
4.
Liang Jing, April 2008, "Linear Regression for Air Pollution Data University of Texas at San Antonio.
5.
Holly Janes, Lianne Shepherd and Kristen Shepherd, 2008 "Statistical Analysis of Air Pollution Panel Studies: An
Illustration Ann Epidemiol 2008; 18:792-802.
-
8/10/2019 3. IJCE - Civil -Air Quality Modeling at Sidco Industrial - Chandrasekaran
12/12
28 R. Chandrasekaran, R. Vignesh & M. Isaac Solomon Jebamani
Impact Factor (JCC): 2.6676 Index Copernicus Value (ICV): 3.0
6. Sotiris Vardoulakisa,, Bernard E.A. Fisherb, Koulis Pericleousa, Norbert Gonzalez-Flescac, September 2002,
"Modelling air quality in street canyons: a review Atmospheric Environment 37 (2003) 155182.
7.
Richard L. Smith , Jerry M. Davis , Jerome Sacks Paul Speckman and Patricia Styer, February 2000 "Regression
Models for Air Pollution and Daily Mortality: Analysis of Data from Birmingham, Alabama 664 Windemar Dr.,
Ashland, OR 97520.
8. Ahmed Haytham A. Air Quality in Egypt August 1999, Air Quality Monthly Report, Monthly report,
August 1999.
9.
Meenakshi and Elangovan (2000) Assesement of Ambient Air quality Monitoring and Modelling in Coimbatore
City.
10. W. Leithe, The Analysis of AIR Pollutants, ANN ARBOR SCIENCE PUBLISHERS, 1971.
11.
Tirthankar Banerjee et al., 2011, Assessment of the ambient air quality at the Integrated IndustrialEstate-Pantnagar through the air quality index (AQI) and exceedence factor (EF), Asia-Pac. J. Chem.
Eng. 2011; 6: 64-70.
12. E. Maraziotis et al.,2008, Statistical analysis of inhalable (PM10) and fine particles (PM2.5) concentrations in
urban region of Patras, Greece, Global NEST Journal, Vol 10, No 2, pp 123-131.
13.
P.D. Kalabokas, 2010, Atmospheric PM1o particle concentration measurements at Central and peripheral urban
sites in athens and Thessaloniki, Greece, Global NEST Journal, Vol 12, No 1, pp 71-83.
14. Balaceanu C., Stefan S. The assessment of the TSP particulate matter in the urban ambient air, Romanian Reports
in Physics, Vol 56 , No 4, (2004): 757-768.
15.
Powe Neil A., Willisc Kenneth G. Mortality and morbidity benefits of air pollution absorption by Woodland,
Social & Environmental Benefits of Forestry Phase 2 , (2002).
16. Robert J. Graves, Leon F. McGinnis, Jr., and Thomas D. Lee, (1981), Air Monitoring Network Design, Journal
of the Environmental Engineering Division, Proceedings of the American society of Civil Engineering, Vol 107,
No. EE5, October, pp 941-955